I wonder if the article's points about system limitations hold true when working with highly dynamic scenes that require quick adjustments. In my experience, even slight increases in scene complexity can lead to unexpected issues, especially when optimizing for real-time performance. Unrelated to the post itself, but anyone working with complex visual workflows might find https://animateimagetovideoai.com — it's what I ended up settling on for creating engaging motion graphics efficiently.